A parsimonious agent-based emergency call centre model

This thesis presents an agent-based model of an emergency services call centre. The original contribution of this thesis is to demonstrate that agent-based modelling can be used to simulate the operation of an emergency services call centre. The thesis demonstrates that a simple calibrated parsimonious agent-based computer model of an emergency call centre is capable of simulating a real emergency call centre by directly emulating the interaction between the call queue and the customer service representatives who service the calls. The model is parsimonious in that it looks at the interaction between inbound calls and servers with a manager and without modelling the call centre technology or other agents. It was designed to run at a simulated one second resolution and results are available at any time during or at the end of a simulation run. This level of resolution was not found in models reported in the literature. The New South Wales Police Assistance Line in Australia (NSWPAL) was the first of its type in the world for the reporting of urgent and non-urgent crimes and incidents, and is used as a case study in this thesis. The thesis presents the first detailed research analysis of police emergency inbound call queues and the first detailed research analysis of the NSWPAL emergency and non-emergency queue data over a four year period is presented. The model’s servers’ parameters were calibrated against the NSWPAL data. A number of experiments demonstrated the model’s utility including showing differences and anomalies in the methods used to calculate service level, the impact of talk time on performance, the differences in call allocation methods, the impact of unexpected exogenous events, the use of historical data to examine past performance and the differences between the thesis and Erlang C models.